This is Jenkins CI/CD demo project. You can create custom workflows to automate your project software life cycle process.
You need to configure workflows using YAML syntax, and save them as workflow files in your repository. Once you've successfully created a YAML workflow file and triggered the workflow - Jenkins parse flow and execute it.
-
Copy
.ci/Jenkinsfile.shlib
to your new github project, under.ci/
-
Copy
.ci/Makefile
to your new github project, under.ci/
-
Create
.ci/job_matrix.yaml
basic workflow file with content:
---
job: ci-demo
registry_host: harbor.mellanox.com
registry_path: /swx-storage/ci-demo
registry_auth: swx-storage
volumes:
- {mountPath: /hpc/local, hostPath: /hpc/local}
- {mountPath: /auto/sw_tools, hostPath: /auto/sw_tools}
- {mountPath: /.autodirect/mtrswgwork, hostPath: /.autodirect/mtrswgwork}
- {mountPath: /.autodirect/sw/release, hostPath: /.autodirect/sw/release}
nfs_volumes:
- {serverAddress: r1, serverPath: /vol/mtrswgwork, mountPath: /.autodirect/mtrswgwork}
- {serverAddress: r3, serverPath: /vol/mlnx_ofed_release_flexcache, mountPath: /auto/sw/release/mlnx_ofed, readOnly: true}
pvc_volumes:
- {claimName: nbu-swx-storage-devops-pvc, mountPath: /mnt/pvc, readOnly: false}
secret_volumes:
- {secretName: 'mellanox-debs-keyring', mountPath: '/mnt/secret'}
empty_volumes:
- {mountPath: /var/home/user/.local/share/containers, memory: false}
kubernetes:
cloud: swx-k8s
runs_on_dockers:
- {file: '.ci/Dockerfile.centos7.7.1908', name: 'centos7-7', tag: 'latest'}
- {file: '.ci/Dockerfile.ubuntu16-4', name: 'ubuntu16-4', tag: 'latest'}
matrix:
axes:
flags:
- '--enable-debug'
- '--prefix=/tmp/install'
arch:
- x86_64
pipeline_start:
run: echo Starting new job
pipeline_stop:
run: echo All done
steps:
- name: Install mofed
run: |
echo Installing driver: ${driver} ...
mofed_installer_exe=/auto/sw/release/mlnx_ofed/MLNX_OFED/mlnx_ofed_install
mofed_installer_opt='--user-space-only --without-fw-update --all -q --skip-unsupported-devices-check'
sudo env build=$driver $mofed_installer_exe $mofed_installer_opt
- name: Configure
run: |
./autogen.sh
./configure $flags
- name: Build
run: make -j 2 all
- name: Install
run: make -j 2 install
-
Copy
.ci/proj_jjb.yaml
to.ci
folder in your project and change github URL to point to your own github project as here. Also change the value of thejjb_proj: 'ci-demo'
to the name you want to give to your job. -
Register new Jenkins project via jenkins-job cli (or create new with UI)
% cd .ci; make jjb
- Trigger run via Jenkins UI
The CI scripts are located in .ci
folder
Jenkins behavior can be controlled by job_matrix.yaml file which has similar syntax/approach as Github actions.
-
jenkinsfile parses .ci/job_matrix*.yaml files
-
the job is bootstrapped and executing according to steps, as defined in yaml file
-
The method works with vanilla jenkins Jenkinsfile
-
The method also works with pipeline as shared library Jenkinsfile.shlib and uses this shlib
-
Basic job definition can be found at
.ci/job_matrix_basic.yaml
-
Advanced job definition is at
.ci/job_matrix.yaml
-
The actual Jenkinsfile that gets executed is here at
.ci/Jenkinsfile.shlib
- Job file can contain array of steps, which are executed sequentially in the docker
- Job file can define own matrix and Jenkinsfile will generate axis combinations to execute in the docker
- Job file can define list of dockerfiles, in
.ci/Dockerfiles.*
that will be added to matrix combinations. - Job file can define if need to build docker from dockerfiles and push resulting image to the registry or just fetch image from registry.
- Job can define different environment variables that will be added to step run environment
- Job can define optional include/exclude filters to select desired matrix dimensions
- Demo contains Jenkins Job Builder config file which loads Jenkins project definition into Jenkins server.
- Jenkins project descibed by file supports following UI actions/parameters:
- boolean - rebuild docker files
- string - Use named Dockerfile (defaul: .ci/Jenkinsfile.shlib)
- string - Configuration file for Job Matrix (can be regex to load multiple), default: job_matrix.yaml
- You can fetch docker image descibed in job_matrix.yaml and run steps in it to mimic Jenkins k8 approach
% cd .ci
% make shell NAME=ubuntu16-4
docker%% cd /scratch
# the step below needed so workspace files (which belongs to linux $USER)
# will be copied (and owned) as Docker user 'jenkins' so it can be modified
# from docker shell
docker%% cp -pr /scratch /tmp/ws
docker%% cd /tmp/ws
docker%% ./autogen.sh && ./configure && make
- You can build docker locally (if it does not exist in registry) as following
% cd .ci
% make build NAME=ubuntu16-4
Scripts located in resources/actions
folder can be invoked
on runtime without needing to store them in project's CI folder.
Please check examples - job_matrix_actions.yaml
Use this action to manage Nexus Repository Manager. This action allows to create/delete/show nexus hosted repositories and also upload and remove packages. Examples:
# show YUM based repository information
nexus.py yum --url http://swx-repos.mtr.labs.mlnx:8081/ --name test_yum_repo --user user --password password --action show
{
"name": "test_yum_repo",
"url": "http://swx-repos.mtr.labs.mlnx:8081/repository/test_yum_repo",
"online": true,
"storage": {
"blobStoreName": "default",
"strictContentTypeValidation": true,
"writePolicy": "allow_once"
},
"cleanup": {
"policyNames": [
"string"
]
},
"yum": {
"repodataDepth": 1,
"deployPolicy": "STRICT"
},
"component": {
"proprietaryComponents": false
},
"format": "yum",
"type": "hosted"
}
# Create YUM repository
nexus.py yum --url http://swx-repos.mtr.labs.mlnx:8081/ --name test_yum_repo --user user --password password --action create
[08/Apr/2021 18:32:11] INFO [root.create_yum_repo:141] Creating hosted yum repository: test_yum_repo
[08/Apr/2021 18:32:11] INFO [root.create_yum_repo:146] Done
# Remove YUM repository
nexus.py yum --url http://swx-repos.mtr.labs.mlnx:8081/ --name test_yum_repo --user user --password password --action delete
[08/Apr/2021 18:31:29] INFO [root.delete_repository:89] Repository has been deleted: test_yum_repo
# Show APT Repository
nexus.py apt --url http://swx-repos.mtr.labs.mlnx:8081/ --name test_apt_repo --user user --password password --action show
{
"name": "test_apt_repo",
"url": "http://swx-repos.mtr.labs.mlnx:8081/repository/test_apt_repo",
"online": true,
"storage": {
"blobStoreName": "default",
"strictContentTypeValidation": true,
"writePolicy": "allow"
},
"cleanup": {
"policyNames": [
"string"
]
},
"apt": {
"distribution": "focal"
},
"aptSigning": null,
"component": {
"proprietaryComponents": true
},
"format": "apt",
"type": "hosted"
}
# Remove APT repository
nexus.py apt --url http://swx-repos.mtr.labs.mlnx:8081/ --name test_apt_repo --user user --password password --action delete
[08/Apr/2021 18:34:09] INFO [root.delete_repository:89] Repository has been deleted: test_apt_repo
# Create APT repository
./nexus.py apt --url http://swx-repos.mtr.labs.mlnx:8081/ --name test_apt_repo --user user --password password --action create --keypair-file /tmp/debs-keyring.priv --distro focal
[08/Apr/2021 18:34:52] INFO [root.create_apt_repo:208] Creating hosted APT repository: test_apt_repo
[08/Apr/2021 18:34:52] INFO [root.create_apt_repo:213] Done
---
# Job name
job: ci-demo
# Specify if containerSelector and agentSelector options per step are mutually
# exclusive (default: true)
step_allow_single_selector: false
# URI to docker registry
registry_host: harbor.mellanox.com
# Path to project`s dockers space under registry
registry_path: /swx-storage/ci-demo
# optional: Path to jnlp containers under $registry_host
# if not set - auto-calculated to $registry_path/..
registry_jnlp_path: /swx-storage
# Credentials (must be defined in Jenkins server configuration) to for access to registry
registry_auth: swx-storage
# k8 cloud name (must be defined in Jenkins server configuration)
kubernetes:
# cloud name to use for all containers
# cloud tag can be specified per specific container in runs_on_containers section
cloud: swx-k8s
# Example how to use k8 node selector to request specific nodes for allocation
nodeSelector: 'beta.kubernetes.io/os=linux'
# optional: enforce limits so that the running container is not allowed to use
# more of that resource than the limit you set. If a Container specifies its own limit,
# but does not specify a request, Kubernetes automatically assigns a request
# that matches the limit.
limits: "{rdma/bw_port_a: 1, hugepages-1Gi: 1024Mi, hugepages-2Mi: 1024Mi, memory: 2048Mi}"
# optional: request additional k8s resources that isn't supported by
# kubernetes-plugin by default
requests: "{hugepages-1Gi: 1024Mi, hugepages-2Mi: 1024Mi, memory: 2048Mi}"
# optional: annotations can be used to attach arbitrary non-identifying metadata to objects
annotations:
- {key: 'k8s.v1.cni.cncf.io/networks', value: 'roce-bw-port1@roce0'}
# optional: container capabilities to add
caps_add: "[ IPC_LOCK, SYS_RESOURCE ]"
# optional: tolerations. Tolerations allow the scheduler to schedule pods with matching taints.
tolerations: "[{key: 'feature.node.kubernetes.io/project', operator: 'Equal', value: 'SPDK', effect: 'NoSchedule'}]"
# optional: can specify jenkins defined credentials and refer/request by credentialsId in step that
# requires it (it's considered usernamePassword by default if 'type' is not specified)
credentials:
- {credentialsId: '311997c9-cc1c-4d5d-8ba2-6eb43ba0a06d', usernameVariable: 'SWX_REPOS_USER', passwordVariable: 'SWX_REPOS_PASS'}
- {credentialsId: 'jenkins-pulp', usernameVariable: 'pulp_usr', passwordVariable: 'pulp_pwd'}
- {credentialsId: 'github-ssh-rsa-key', 'keyFileVariable': 'SSH_KEY_FILE', type: 'sshUserPrivateKey'}
- {credentialsId: 'secret-file', 'variable': 'SECRET_FILE', type: 'file'}
# optional: for multi-arch k8 support, can define arch-specific nodeSelectors
# and jnlpImage locations
arch_table:
x86_64:
nodeSelector: 'kubernetes.io/arch=amd64'
jnlpImage: 'jenkins/inbound-agent:latest'
aarch64:
nodeSelector: 'kubernetes.io/arch=arm64'
jnlpImage: '${registry_host}/${registry_jnlp_path}/jenkins-arm-agent-jnlp:latest'
# volumes to map into dockers
volumes:
- {mountPath: /hpc/local, hostPath: /hpc/local}
- {mountPath: /auto/sw_tools, hostPath: /auto/sw_tools}
- {mountPath: /.autodirect/mtrswgwork, hostPath: /.autodirect/mtrswgwork}
- {mountPath: /.autodirect/sw/release, hostPath: /.autodirect/sw/release}
# NFS volumes to map into dockers
nfs_volumes:
- {serverAddress: r1, serverPath: /vol/mtrswgwork, mountPath: /.autodirect/mtrswgwork}
- {serverAddress: r3, serverPath: /vol/mlnx_ofed_release_flexcache, mountPath: /auto/sw/release/mlnx_ofed, readOnly: true}
# PersistentVolumeClaim volumes to map into containers
pvc_volumes:
- {claimName: nbu-swx-storage-devops-pvc, mountPath: /mnt/pvc, readOnly: false}
# secretVolume volumes to map into containers
secret_volumes:
- {secretName: 'mellanox-debs-keyring', mountPath: '/mnt/secret'}
# emptyDir volumes to map into containers
# memory flag creates volumes in RAM instead of Disk (size of RAM depends on memory limits/requests)
empty_volumes:
- {mountPath: /var/home/user/.local/share/containers, memory: false}
# environment varibles to insert into Job shell environment, can be referenced from steps
# or user-scripts or shell commands.
# If you want more detailed output in logs, add the following environment variable:
# `DEBUG: true`
env:
mofed_installer_exe: /.autodirect/sw/release/mlnx_ofed/MLNX_OFED/mlnx_ofed_install
mofed_installer_opt: --user-space-only --without-fw-update --all -q --skip-unsupported-devices-check
# default variables and values that can be used in yaml file
defaults:
var1: value1
var2: value2
# list of dockers to use for the job, `file` key is optional, if defined but docker image
# does not exist in registry.
# `arch` key is optional, it defaults to `x86_64` unless specified otherwise.
# image will be created during 1st invocation or if file was modified
# runs_on_dockers list can contain use-defined keys as well
# category:tools has special meaning - it will run for steps, that explicitly request it in the
# step`s containerSelector key.
# The use-case is as following: if some step requires special container with pre-installed toolchain (clang?)
runs_on_dockers:
- {file: '.ci/Dockerfile.centos7.7.1908', name: 'centos7-7', tag: 'latest', category: 'tool'}
- {file: '.ci/Dockerfile.ubuntu16-4', name: 'ubuntu16-4', tag: 'latest'}
# list of jenkins agents labels to run on (optional)
runs_on_agents:
- nodeLabel: '(dockerserver || docker) && x86_64'
- nodeLabel: 'hpc-test-node-inbox'
# user-defined matrix to run tests, "steps" will be executed for every dimension of matrix.
# Can contain any use-defined dimensions
# Docker list will be added automatically to dimensions list
# foreach image in dockers_list
# foreach driver in drivers_list
# foreach cuda in cuda_list
# foreach arch in arch_list
# run steps
# done
# done
# done
# done
#
# Note that the matrix should ALWAYS include the `arch` axis.
# Other variables are optional and should be set as per your needs.
matrix:
axes:
driver:
- MLNX_OFED_LINUX-4.9-0.1.8.0
- MLNX_OFED_LINUX-5.1-1.0.0.0
cuda:
- dev/cuda9.2
arch:
- x86_64
# include only dimensions as below. Exclude has same syntax. Only either include or exclude can be used.
# all keywords in include/exclude command are optional - if all provided keys
# match - the dimension will be include/excluded
include:
- {arch: x86_64, cuda: dev/cuda11.0, driver: MLNX_OFED_LINUX-4.9-0.1.8.0, name: ubuntu16-4}
- {arch: x86_64, cuda: dev/cuda9.2, driver: MLNX_OFED_LINUX-4.9-0.1.8.0, name: ubuntu16-4}
# Steps can contain any number of name/run sections and all will be executed
# every matrix dimension will run in parallel with all other dimensions
# steps itself has sequential execution
# NOTE:
# shell environment variable representing dimension values will be inserted automatically and
# can be used run section (see below)
# $name represents current value for docker name
# Also $driver $cuda,$arch env vars are available and can be used from shell
# $variant represents current axis serial number (relative to docker loop)
steps:
- name: Coverity scan
# `shell` can be type of action, it will run script located in ci-demo/vars/actions/scriptname.sh
# with parameters
# defined by `args` key.
shell: action
# dynamicAction is pre-defined action defined at https://github.com/Mellanox/ci-demo/blob/master/vars/dynamicAction.groovy
# dynamicAction will execute ci-demo/vars/actions/$args[0] and will pass $args[1..] to it as command line arguments
run: dynamicAction
# step can specify containerSelector filter to apply on `runs_on_dockers` section.
# make sure to quote the category.
# `variant` is built-in variable, available for every axis of the run and represents serial number for
# execution of matrix dimension
# selector can be regex
containerSelector: '{category:"tool", variant:1}'
args:
- "--pre_script './autogen.sh;./configure;make -j 3 clean'"
- "--build_script 'make -j 3'"
- "--ignore_files 'devx googletest tests'"
archiveArtifacts: 'cov.log'
# run this step in parallel with others
# each non-parallel step is a barrier for previous group of parallel steps
parallel: true
- name: Check package
# use 'jenkins-pulp' credentials in this steps
credentialsId: 'jenkins-pulp'
# can set shell per step or globally
shell: '!/bin/bash -xeEl'
run: cuda=$cuda .ci/check_package.sh
parallel: true
- name: Run tests
# multiple credentials can be used as a list in one step
credentialsId:
- 'jenkins-pulp'
- 'github-ssh-rsa-key'
- 'secret-file'
run: cuda=$cuda .ci/runtests.sh
# define shell command(s) to run if step fails
onfail: |
echo step execution step failed
touch myfile.html
touch step_failed.log
# define shell command to run always, regardless if "run" step passed or failed
always: env > always_env.txt
# define artifacts to collect for specific step
archiveArtifacts-onfail: 'step_failed.log'
archiveArtifacts: 'always_env.txt'
# Publish HTML on jenkins page for given run
publishHTML:
reportDir: '.'
reportFiles: 'myfile.html'
reportName: 'Test'
# define raw xml results to collect for specific step
archiveJunit: 'test-results.xml'
# define TAP results to collect for specific step (see jenkins TAP plugin)
archiveTap: '**/*.tap'
# executed once, before job starts its steps
pipeline_start:
run: echo Starting new job
# executed once, after steps are done
pipeline_stop:
run: echo All done
# executed before each container image build phase
# user can run any script to modify container image content
# also, can define 'on_image_build: script' key in runs_on_containers
# section , per specific containers
pipeline_on_image_build:
run: echo Building image
# List of artifacts to attach to Jenkins results page for build
archiveArtifacts: config.log
# List of raw xml results to attach to Jenkins results page for build
archiveJunit: 'myproject/target/test-reports/*.xml'
# Fail job is one of the steps fails or continue
failFast: false
# Execute parallel job in batches (default 10 jobs in the air), to prevent overload k8
# with large amount of parallel jobs
batchSize: 2
# Job timeout - fail job if it runs more than specified amount of minutes (default is 90 minutes)
timeout_minutes: 60
# Customize name of the parallel subtask as appears in Jenkins UI, according to the template below
# can use variable names from `axis` part of the `matrix` config section
# also can use variable names from `runs_on_dockers` config section.
# `${name}` comes from `runs_on_dockers` section
# `${axis_index}` is built-in variable representing axis serial number
taskName: '${name}/${axis_index}'